Copyright Notice:

The documents distributed by this server have been provided by the contributing authors as a means to ensure timely dissemination of scholarly and technical work on a noncommercial basis. Copyright and all rights therein are maintained by the authors or by other copyright holders, notwithstanding that they have offered their works here electronically. It is understood that all persons copying this information will adhere to the terms and constraints invoked by each author's copyright. These works may not be reposted without the explicit permission of the copyright holder.

Publications of SPCL

P. Mazumder, A. Calotoiu, M. Copik, Y. Koray Budanaz, S. Chandran, T. Hoefler:

 What Bloats Your Floats? Right-Sizing Numerical Precision for Scientific Computing

(ACM Trans. Reconfigurable Technol. Syst.. Vol , Nr. , In , presented in New York, NY, USA, pages , ACM, ISSN: 1936-7406, ISBN: , Mar. 2026, Just Accepted )

Publisher Reference

Abstract

While FP64 (binary64) remains the standard representation of real numbers in scientific computing, hardware trends increasingly favor low-precision formats optimized for AI workloads. This shift prioritizes throughput and energy efficiency over full-precision capabilities. We challenge the necessity of high-precision arithmetic in PDE-governed simulations, where spatial and temporal discretization errors introduce a physical noise floor that masks the lower-order bits of the FP64 format, rendering them irrelevant. We propose a criterion based on the Signal-to-Noise Ratio (SNR) with respect to spatial and temporal refinements to evaluate the impact of safe precision reduction on scientific computations. Finally, we present an automated dataflow-centric workflow for targeted lowering within mixed-precision subgraphs in complex scientific applications. Large-scale evaluations of weather and climate applications show that strategic precision reduction on GPUs achieves up to 1.8× speedup without compromising physical fidelity.

Documents

download article:
download slides:
 

BibTeX

@article{pratyai2026floats,
  author={Pratyai Mazumder and Alexandru Calotoiu and Marcin Copik and Yakup Koray Budanaz and Shriram Chandran and Torsten Hoefler},
  title={{What Bloats Your Floats? Right-Sizing Numerical Precision for Scientific Computing}},
  journal={ACM Trans. Reconfigurable Technol. Syst.},
  institution={ETH Zurich},
  year={2026},
  month={03},
  pages={},
  volume={},
  number={},
  booktitle={},
  location={New York, NY, USA},
  publisher={ACM},
  issn={1936-7406},
  isbn={},
  note={Just Accepted},
  doi={0.1145/3799896},
}